Main     Guide 9Y40 by Richard Pavlicek    

About the Stats

Partnership stats are obtained from vugraph archives of four annual events: Vanderbilt Knockout Teams, Spingold Knockout Teams, United States Bridge Championship and World Team Championship. From a U.S. perspective, virtually all would consider these the four most difficult and prestigious IMP team events on the calendar. For three of the events (all but U.S. Championship) the same is true worldwide, as their fields are cluttered with the world’s best players. Therefore, stats from these events offer convincing evidence for a creditable global ranking system.

Archives do not include all boards played in each event but only those recorded for vugraph. This raises concern about their validity, but in my view it actually improves the data collection. When a match is selected for vugraph, it is usually because both teams are strong or closely contended, so the stats are meaningful. In contrast, when a match is not selected, it is usually because neither team is a contender or because a strong team is mismatched against a weak team. In the latter case, the good stats of the strong team would be tainted and only serve to pad their rating if recorded. Consequently, a high ranking is earned, not gifted.*

*Vugraph organizers occasionally succumb to poor match selection, and the 2012 World Championship was a cause celebre. Perhaps in a frenzy to appease Italian rooters, qualifying-round matches of Italy vs. Kenya, Italy vs. Bermuda, and Italy vs. Trinidad/Tobago managed to reach the vugraph screen. Was the object to see if Italy could run the table? (Just about, as the IMP scores were 87-0, 88-2 and 101-6, respectively.) Sorry, no freebies. Those “matches” have been removed from this archive.

One method of partnership ranking is average IMPs per board, but this is always an evaluation of two pairs (current teammates) rather than the effectiveness of a single pair. The method I use is average IMPs to par, a benchmark that is independent of a pair’s teammates.

How it works

For each board, a par score is determined based on perfect bidding and double-dummy play all-around. Basically, this is the highest scoring makable contract, unless the other side has a cheaper doubled sacrifice. For example, if the best makable contract is four spades by East (none vulnerable), par is presumed to be 420 for East-West; but if North or South can win nine tricks in clubs, par is 300 for East-West (five clubs doubled down two). In the rare event that neither side can make any contract, par is zero (a perfect passout).

A pair’s actual score on each board is compared to the par score, and the net difference is converted to IMPs (plus or minus). The total is then divided by the number of boards played to find the average. A plus average is good; a minus average is not good. Note that this is a zero-sum method; i.e., average IMPs to par for all pairs adds to zero.

In practice, a partnership will almost never be plus (to par) on boards declared. This is obvious if you think about it, because some best contracts are too elusive for anyone to reach, and even a reachable best contract may require double-dummy play to make. Conversely, a partnership will almost always be plus on boards defended. This quirk means that the side with more HCP on a deal has the more difficult task to attain par. Of course, the HCP distribution evens out over a significant number of boards, so the quirk has virtually no effect on a pair’s rating.

Is it completely fair?

No. Is anything? One factor not considered is difficulty of schedule, i.e., the strength of each pair’s opponents at the table. While this tends to balance out, I can think of some scenarios for bias. When two sponsored teams face off, the client pairs (those with the sponsor) tend to play against each other, which softens the turf for both client pairs; so they may be overrated here. Conversely, I’ve noticed that some pairs (Levin-Weinstein come to mind) usually take on the opposing team’s strongest pair; so they may be underrated here.

I’ve thought about incorporating a degree-of-difficulty factor into each encounter; but it would be complicated to implement, and any method used would itself be subject to doubts of fairness. Therefore, I junked the idea. One has to accept that in bridge, as in life, luck will always be an element.


© 2011 Richard Pavlicek